Computer-aided identification of renal corpuscle elements in RGB and HLS color images

The qualitative and quantitative changes in the structure of the renal corpuscle allow the assessment of the glomerular involvement degree, defining each type of glomerulonephritis. Recently, several works dedicated to glomerulonephritis focus on the quantitative investigation, achieved by digital (numerical) image processing, where the identification of the interest elements is a key stage.

The aim of the study was the elaboration of a technique for constructing reliable color models in the RGB and HLS spaces for each element of the structure of the renal corpuscle.

We analyzed images captured from trichrome stained fragments of renal biopsy. Each image contained a single renal corpuscle. The identification procedure used the software tools specialized in color image analysis provided by the Zeiss KS400 environment.

There were defined specific chromatic domains in the RGB and HLS space for the identification of the following elements: the Bowman space and the glomerular capillaries (white colored regions), the nuclei (brown colored regions), the mesangium (green colored regions).

In order to get a relevant picture of the role played by the decisions made in the definition of the chromatic domains, we performed two tests for the modification of the chromatic domain allocated to the mesangium in the RGB and HLS space, respectively.

The procedures proposed for the identification of the RGB and HLS color models can successfully replace the manual delineation of the renal corpuscle elements. Such models represent an important step in our research visualizing the future development of computer programs for automated measurements of renal corpuscle elements.